This poster presents research results from a recent survey studying metadata workflows. The survey was distributed to DataNet Federation Consortium researchers and collaborators asking participants about their organizational practices involving metadata creation. Data management best practices recommend that data documentation happens at the very beginning of the research project, before data collection. However, these results indicate that more scientific metadata is created during or after the data collection process than before, and that few researchers take advantage of automated metadata generation workflows. Data curators, librarians, and archivists (or their automated systems) can assist researchers by intervening earlier in the data life cycle in order to produce higher-quality metadata and ensure long-term preservation.